Meet Deep Thunder: IBM’s next step in the automation of forecasting

Five months after acquiring The Weather Company, IBM continues to move toward automated forecasting. (credit: IBM)

Until recently weather forecasting was a fairly straightforward process. Scientists and meteorologists with a government agency developed computer forecast models, collected data about current weather conditions, input that data into their models, and then ran them on government hardware. A TV forecaster would next review the output of these models and give you the weather during the 6 and 10 o’clock news.

But in recently the private US weather industry, valued at between $3 billion and $6 billion, has gone far beyond this traditional method of forecasting. Because the National Weather Service is federally funded, the agency makes both the basic code of its model, as well as the raw output, available to both research and commercial entities. Companies have taken the government’s models and “added value” for consumer and business customers.

In late January of this year, IBM finalized its acquisition of The Weather Company, buying all of its assets except for The Weather Channel television network. Both IBM and The Weather Company had been working separately with one of the government’s most popular models, the WRF, or Weather Research and Forecasting Model. Developed in the late 1990s, the WRF is tuned to provide more accurate local forecasts rather than predicting conditions across the globe. (Other companies, such as Panasonic, have developed their own global models based upon the government's code).

Read 7 remaining paragraphs | Comments